THE USE OF ARTIFICIAL INTELLIGENCE IN BASIC EDUCATION: ETHICAL CHALLENGES AND PEDAGOGICAL POSSIBILITIES

Abstract

This integrative literature review aimed to analyze the dialectical interaction between transformative pedagogical possibilities and critical ethical challenges associated with the integration of Artificial Intelligence (AI) in Basic Education (K-12), synthesizing a comprehensive framework for responsible educational practice. The methodology consisted of an integrative review of 46 scientific studies (2020-2026), selected after segmented searches in the Scopus database on pedagogical and ethical dimensions, followed by thematic analysis. The results demonstrated that AI offers significant pedagogical possibilities, centered on the personalization of learning at scale, the development of critical and creative thinking (via pedagogies such as Design Fiction and the CEDE model), the expansion of AI literacy, support for STEM education, and self-directed teacher professional development. However, these potentialities are intrinsically strained by a complex network of ethical and operational challenges, including risks of algorithmic bias and discrimination, violations of children's data privacy, lack of transparency and accountability, widening digital divide, and, crucially, inadequate teacher preparation. The discussion highlighted that the integration of AI transcends the technical dimension, configuring itself as a socio-technical, political, and pedagogical process that requires critical mediation. It was concluded that the realization of pedagogical possibilities depends on overcoming ethical challenges through a new formative pact, structured on three pillars: 1) the reconversion of teacher training to an Intelligent-TPACK model that integrates technical, pedagogical, and ethical reflection; 2) the transversal incorporation of critical AI literacy into the K-12 curriculum, forming digital citizens; and 3) the implementation of institutional ethical governance frameworks (such as GEAI and the HCAI Block Model) that guarantee privacy by design, data sovereignty, and human oversight (human-in-the-loop). The path to responsible adoption therefore lies in a stance of "cautious optimism," which subordinates technology to a humanistic educational project, investing massively in the training of human actors as central agents of this transformation.